CN-121998750-A - Training method, device, equipment and storage medium of business handling evaluation model

Abstract

The application discloses a training method, device and equipment of a business handling evaluation model and a storage medium, which can be applied to the field of financial science and technology. The method comprises the steps of obtaining historical business handling data of target banking business, constructing business sample data based on the historical business handling data, performing first model training on a machine learning model constructed in advance based on the business sample data to obtain candidate business handling evaluation models, respectively performing business handling prediction on different business dimension feature combinations in the historical business handling data by adopting the candidate business handling evaluation models to obtain candidate prediction business handling labels, determining feature importance of different business dimension features in the historical business handling data based on the prediction business handling labels, and performing second model training on the candidate business handling evaluation models based on the feature importance and the historical business handling data to obtain the target business handling evaluation models. Through the technical scheme, the model performance and the prediction precision are improved.

Inventors

• WANG TONG
• HOU CHENGLONG
• FAN WENJING

Assignees

• 中国工商银行股份有限公司

Dates

Publication Date

20260508

Application Date

20260205

Claims (10)

1. 1. The training method of the business handling evaluation model is characterized by comprising the following steps of: acquiring historical business handling data of a target banking business, and constructing business sample data based on the historical business handling data, wherein the business sample data comprises a real business handling label; performing first model training on a pre-constructed machine learning model based on the service sample data until a first model training ending condition is met, so as to obtain a candidate service handling evaluation model; Respectively carrying out service handling prediction on different service dimension feature combinations in the historical service handling data by adopting the candidate service handling evaluation model to obtain candidate predicted service handling labels, and determining feature importance of different service dimension features in the historical service handling data based on the predicted service handling labels; And performing second model training on the candidate business handling evaluation model based on the feature importance and the historical business handling data until a second model training ending condition is met, so as to obtain a target business handling evaluation model.
2. 2. The method according to claim 1, wherein the employing the candidate business transaction evaluation model to respectively predict the business transaction for the different business dimension feature combinations in the historical business transaction data to obtain candidate predicted business transaction tags, and determining the feature importance of the different business dimension features in the historical business transaction data based on the predicted business transaction tags comprises: Inputting the historical business handling data into a candidate business handling evaluation model for feature extraction, and generating a business dimension feature set and at least one business dimension feature subset, wherein the business dimension feature combinations existing in different business dimension feature subsets are different; respectively carrying out service handling prediction on the service dimension feature set and the service dimension feature subset to obtain respective corresponding candidate predicted service handling labels; And determining the feature importance of each service dimension feature according to the service dimension feature set, the service dimension feature subset and the candidate predicted service handling labels corresponding to the service dimension feature set and the service dimension feature subset.
3. 3. The method of claim 1, wherein the service dimension feature subset is a subset of a service dimension feature set, the number of feature categories present in different service dimension feature subsets is the same, and the difference in the number of feature categories between the service dimension feature subset and the service dimension feature set is 1.
4. 4. The method of claim 1, wherein the performing a second model training on the candidate business transaction evaluation model based on the feature importance and the historical business transaction data comprises: Inputting the historical business handling data into the candidate business handling evaluation model, and extracting features of the historical business handling data to generate a business dimension feature set; Performing feature fusion on each service dimension feature in the service dimension feature set and the corresponding feature importance degree respectively to generate a target fusion feature set; And carrying out service handling prediction by adopting the target fusion feature set to obtain a target prediction service handling label, and carrying out second model training on the candidate service handling evaluation model according to the target prediction service handling label and the real service handling label.
5. 5. The method of claim 1, further comprising, prior to first model training a pre-built machine learning model based on the business sample data: and carrying out parameter optimization on the metadata model structure parameters in the machine learning model by adopting a preset model parameter optimization algorithm.
6. 6. The method of claim 1, wherein the objective loss function in the first model training and the second model training is determined based on a deviation between a real business transaction tag and a predictive business transaction tag, and regularization parameters characterizing a model structure of a machine learning model.
7. 7. A training device for a business transaction evaluation model, comprising: The system comprises a sample generation module, a target bank business processing module and a target bank business processing module, wherein the sample generation module is used for acquiring historical business processing data of the target bank business and constructing business sample data based on the historical business processing data, and the business sample data comprises a real business processing tag; The first training module is used for carrying out first model training on a pre-constructed machine learning model based on the service sample data until a first model training ending condition is met, so as to obtain a candidate service handling evaluation model; The feature importance module is used for respectively carrying out service handling prediction on different service dimension feature combinations in the historical service handling data by adopting the candidate service handling evaluation model to obtain candidate predicted service handling labels, and determining the feature importance of different service dimension features in the historical service handling data based on the predicted service handling labels; And the second training module is used for carrying out second model training on the candidate business handling evaluation model based on the feature importance and the historical business handling data until a second model training ending condition is met, so as to obtain a target business handling evaluation model.
8. 8. An electronic device, comprising: One or more processors; a memory for storing one or more programs; The one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of training a business transaction evaluation model according to any one of claims 1-6.
9. 9. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements a method for training a business transaction evaluation model according to any one of claims 1-6.
10. 10. A computer program product comprising a computer program which, when executed by a processor, implements a method of training a business transaction evaluation model according to any one of claims 1-6.

Description

Training method, device, equipment and storage medium of business handling evaluation model Technical Field The embodiment of the application relates to the technical field of computers, and can be applied to the field of financial science and technology, in particular to a training method, device, equipment and storage medium of a business handling evaluation model. Background At present, most of the existing customer groups of banks have financial requirements, and the customers with the financial requirements are determined whether to be really suitable for transacting financial services, so that better financial services are provided for the customers, matched financial products and services are provided for the customers, the viscosity of the customers can be obviously improved, the customer experience can be optimized, and the core competitiveness of the commercial banks is improved. Therefore, there is a need for a method for evaluating banking transactions. Disclosure of Invention The application provides a training method, a device, equipment and a storage medium of a business handling evaluation model, which are used for improving the adaptation degree of banking business handling and clients, providing better service for the clients and improving the viscosity of the users. According to one aspect of the present application, there is provided a training method of a business transaction evaluation model, the method comprising: acquiring historical business handling data of a target banking business, and constructing business sample data based on the historical business handling data, wherein the business sample data comprises a real business handling label; performing first model training on a pre-constructed machine learning model based on the service sample data until a first model training ending condition is met, so as to obtain a candidate service handling evaluation model; Respectively carrying out service handling prediction on different service dimension feature combinations in the historical service handling data by adopting the candidate service handling evaluation model to obtain candidate predicted service handling labels, and determining feature importance of different service dimension features in the historical service handling data based on the predicted service handling labels; And performing second model training on the candidate business handling evaluation model based on the feature importance and the historical business handling data until a second model training ending condition is met, so as to obtain a target business handling evaluation model. According to another aspect of the present application, there is provided a training apparatus for a business transaction evaluation model, the apparatus comprising: The system comprises a sample generation module, a target bank business processing module and a target bank business processing module, wherein the sample generation module is used for acquiring historical business processing data of the target bank business and constructing business sample data based on the historical business processing data, and the business sample data comprises a real business processing tag; The first training module is used for carrying out first model training on a pre-constructed machine learning model based on the service sample data until a first model training ending condition is met, so as to obtain a candidate service handling evaluation model; The feature importance module is used for respectively carrying out service handling prediction on different service dimension feature combinations in the historical service handling data by adopting the candidate service handling evaluation model to obtain candidate predicted service handling labels, and determining the feature importance of different service dimension features in the historical service handling data based on the predicted service handling labels; And the second training module is used for carrying out second model training on the candidate business handling evaluation model based on the feature importance and the historical business handling data until a second model training ending condition is met, so as to obtain a target business handling evaluation model. According to another aspect of the present application, there is provided an electronic apparatus including: One or more processors; a memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement any one of the training methods for the business transaction evaluation model provided by the embodiment of the present application. According to another aspect of the present application, there is provided a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a training method for any one of the business transaction evaluation models provided by the embodiments of the presen